I recently downloaded REW to measure amplifier distortion and speaker FR. I'm using a Behringer UCA202, a pretty inexpensive external ADC/DAC combo.
My first measurements were just some characterizations of the UCA202's baseline noise and my DDS-based signal generator.
Baseline noise:
There are little blips at 1 and 2 KHz, perhaps some stray pickup from my crappy signal generator:
No way I'm using this thing for distortion measurements! Fortunately the output side of the UCA202 is pretty decent. The spectrum below was taken with my JLH-like headphone amplifier:
The peak at 1KHz is set to ~-8dbV because my dummy-load circuit has two 10K resistors to prevent damaging the ADC input. So it attenuates the signal by 8dBV (my measurements determined that the UCA's input impedance is 12K). So the HA's output actually is running at 0 dBV. The first harmonic's amplitude then is actually about -92dBV. Not too bad considering the design's relatively-low OLG. I'm happy about that.
I'm particularly happy about the absence of power line noise in the measurement. Part of the reason may be that I'm using an isolated USB adapter. It plugs into the laptop's USB socket and transfers power and data via a transformer-coupled power supply and isolated data lines -- there's no possibility of a ground loop. The power supply has to be a switcher but seems to be fairly quiet, as shown by my baseline spectrum.
One "gotcha" I discovered while doing the HA distortion measurements has to do with the UCA202's headphone output. I discovered that if it is turned ON there was enough internal crosstalk to cause the amplifier to oscillate. Those limiting resistors paid for themselves right then and there! I'm using two 1/4W parallel-connected 100 ohm resistors to get 50 ohms, close to my headphones' impedance and they got VERY hot, very quickly when the HA was oscillating.
I also have an SAE MKIII C/M that I want to check out. The SM indicates that the output stage's idle current is adjusted so the THD is .03 - .04% @ 1 watt. REW will be handy for that, too.
My first measurements were just some characterizations of the UCA202's baseline noise and my DDS-based signal generator.
Baseline noise:
There are little blips at 1 and 2 KHz, perhaps some stray pickup from my crappy signal generator:
No way I'm using this thing for distortion measurements! Fortunately the output side of the UCA202 is pretty decent. The spectrum below was taken with my JLH-like headphone amplifier:
The peak at 1KHz is set to ~-8dbV because my dummy-load circuit has two 10K resistors to prevent damaging the ADC input. So it attenuates the signal by 8dBV (my measurements determined that the UCA's input impedance is 12K). So the HA's output actually is running at 0 dBV. The first harmonic's amplitude then is actually about -92dBV. Not too bad considering the design's relatively-low OLG. I'm happy about that.
I'm particularly happy about the absence of power line noise in the measurement. Part of the reason may be that I'm using an isolated USB adapter. It plugs into the laptop's USB socket and transfers power and data via a transformer-coupled power supply and isolated data lines -- there's no possibility of a ground loop. The power supply has to be a switcher but seems to be fairly quiet, as shown by my baseline spectrum.
One "gotcha" I discovered while doing the HA distortion measurements has to do with the UCA202's headphone output. I discovered that if it is turned ON there was enough internal crosstalk to cause the amplifier to oscillate. Those limiting resistors paid for themselves right then and there! I'm using two 1/4W parallel-connected 100 ohm resistors to get 50 ohms, close to my headphones' impedance and they got VERY hot, very quickly when the HA was oscillating.
I also have an SAE MKIII C/M that I want to check out. The SM indicates that the output stage's idle current is adjusted so the THD is .03 - .04% @ 1 watt. REW will be handy for that, too.
I have referred to my "JLH-like" amplifier design a number of times but haven't shared exactly what differentiates it from the original. So here it is:
The main differences are that the input transistors' emitter resistors are terminated in +/- 4V rather than the full supply rails ( they're produced by a pair of on-board LM317/337 voltage regulators). This allowed me to significantly drop the emitter resistor values, which increased the OLG. In addition to that, I didn't see any reason to use the original design's rather low-value collector-load resistors because the auto-bias feature takes care of any excess current. That also increased the OLG. The 71.5K resistors can actually be eliminated without causing much of a problem. Finally, I added two transistors that converted the output stage into a Darlington-pair. That also increased the OLG. Other than those mods, it's vintage JLH 🙂
While the simulations didn't show any tendency to oscillate, the actual circuit turned out to be marginally stable. Under some conditions it DID oscillate, that's why I added the Miller capacitors. In addition to that, I finally had to insert a series resistor in the input line. Called it a base stopper, damping resistor, whatever terminology that's currently in use. The combination of the high Ft transistors I used and higher OLG did turn out to be a little more challenging in terms of achieving circuit stability.
BTW, Q13/14 are there to explore some ways to reduce the amp's turn-on transient. Due to the 100uF capacitors in the auto-bias control loop the initial idle current in the output transistors is pretty high. It also produces a substantial "thump" if the headphones are plugged in when it's turned on.
The main differences are that the input transistors' emitter resistors are terminated in +/- 4V rather than the full supply rails ( they're produced by a pair of on-board LM317/337 voltage regulators). This allowed me to significantly drop the emitter resistor values, which increased the OLG. In addition to that, I didn't see any reason to use the original design's rather low-value collector-load resistors because the auto-bias feature takes care of any excess current. That also increased the OLG. The 71.5K resistors can actually be eliminated without causing much of a problem. Finally, I added two transistors that converted the output stage into a Darlington-pair. That also increased the OLG. Other than those mods, it's vintage JLH 🙂
While the simulations didn't show any tendency to oscillate, the actual circuit turned out to be marginally stable. Under some conditions it DID oscillate, that's why I added the Miller capacitors. In addition to that, I finally had to insert a series resistor in the input line. Called it a base stopper, damping resistor, whatever terminology that's currently in use. The combination of the high Ft transistors I used and higher OLG did turn out to be a little more challenging in terms of achieving circuit stability.
BTW, Q13/14 are there to explore some ways to reduce the amp's turn-on transient. Due to the 100uF capacitors in the auto-bias control loop the initial idle current in the output transistors is pretty high. It also produces a substantial "thump" if the headphones are plugged in when it's turned on.